Re: 30apr00 bugs: spaces in attributs, empty ALT attributes

On 3 May 2000, at 14:37, Sebastian Lange wrote:

> with $Message = '<FONT FaCe="Comic Sans MS">test</A>', $tidiedMessage will be:
> <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 3.2//EN">
> <HTML>
> <HEAD>
> <TITLE></TITLE>
> </HEAD>
> <BODY>
> <FONT FACE="Comic" SANS="" MS="">test</FONT>
> </BODY>
> </HTML>
> 
> The irrelevant lines (doctype to body, /body to /html) are then 
> automatically removed by my perl script, which leaves $tidiedMessage to be 
> '<FONT FACE="Comic" SANS="" MS="">test</FONT>'.
> Having this tidied again, turns $tidiedMessage into '<FONT FACE="Comic" 
> SANS="MS=">test</FONT>' which then stays like that on subsequent tidy attempts.

This looks to me like a bug in lexical analysis. Or possibly two 
(inter-related?) bugs.

1. Something is trying to allow a double-quote-delimited string to 
include paired double-quotes to represent a "real" double-quote. IOW, 
you want a value of xyz"abc, so you code it as "xyz""abc". But it's 
got it wrong and is treating "" (null string) as an attempt at paired 
*internal* double-quotes.

2. Something (else?) is treating an embedded space as terminating a 
double-quoted value.

So:
  Face="Comic Sans MS"
gets tokenised as 
  Face
  =
  "Comic
  Sans
  MS"
and the 'Sans' and 'MS' are then treated as empty-value keywords, so 
"corrected" to 'kwd=""' form, yielding
  Face="Comic"
  Sans=""
  MS=""
as 3 'kwd=value' sets.

Then on the second pass, it gets tokenised as
  Face
  =
  "Comic"
  Sans
  =
  " MS="
(note double-doublequotes reduced to single-doublequotes)
and then output as 2 kwd=value pairs
  Face="Comic"
  Sans="MS="

Does this point you in the right direction? Sorry I can't go into the 
code myself ...

Received on Wednesday, 3 May 2000 10:04:38 UTC